[v2.0] Move psk store to user #237
Conversation
streams/src/message/keyload.rs
Outdated
| fork.mask(&mut psk_id)?; | ||
|
|
||
| let mut masked_key = [0u8; KEY_SIZE]; | ||
| if let Some(psk) = keyload.psk_store.get(&psk_id) { |
There was a problem hiding this comment.
we could check here if the key has not yet been decrypted (key.is_none()), in order to avoid decrypting it multiple times unnecessarily
| #[cfg(feature = "did")] | ||
| Identifier::DID(did) => { | ||
| let oneof = Uint8::new(2); | ||
| let oneof = Uint8::new(1); |
There was a problem hiding this comment.
I guess this is fine now since were doing a large rewrite, but we should keep in mind in the future, that we should not change these numbers
| self.send_keyload( | ||
| link_to, | ||
| // Alas, must collect to release the &self immutable borrow | ||
| self.subscribers().map(Permissioned::Read).collect::<Vec<_>>(), | ||
| subscribers, | ||
| psks, | ||
| ) | ||
| .await | ||
| } | ||
|
|
||
| pub async fn send_keyload_for_all_rw(&mut self, link_to: MsgId) -> Result<SendResponse<TSR>> { |
There was a problem hiding this comment.
hm im not sure how this method passed review earlier. Did we get requested to add this method, or are we sure this will be needed in many cases? If thats the case, should we change the default? Alternatively, i think a way to specify the "Permisison type" per identifier is better. (not in this PR, but should be given thought. (I am referring to timed permissions)
There was a problem hiding this comment.
I think it's worth addressing when we get to the high level API rewrite, because defining the Permissions more directly will make more sense semantically once send_keyload assumes the new change_permissions() definition
|
@kwek20 #241 Arnau made this issue after this conversation: #237 (comment). What's your opinion? |
Description of change
Moves the
psksHashMap out ofKeyStoreand intoUserdirectly.Permissioned::Readto each of them when masking keyType of change
How the change has been tested